Dark Matter Photography
A collection of my photographs and slides.
Sp@m C@tcher
"Your last line of defense against SPAM"
Research Central
My personal site
Office of the Privacy Commissioner of Canada
Collection of essays on DPI (Deep Packet Inspection)
Domain Crawler
[ Thanks to Mary ]
NOTE: This article first appeared in 'Content London', which is no longer published.
May / June 1998 Vol. 2 #5
How many of you can remember the Internet BW (Before Web)? Think back now, it was e-mail, ftp, gopher, and telnet. Remember those? Remember how FAST the Internet was? Whew! Now that was cookin' with gas!.
The topic of this installment is something I hope not too many of you have heard of - who wants to read about what you already know. I can almost guarantee though that all of you have run into that frustrating experience of surfing the Web, only to find that most of your online time is spent waiting for connections to remote servers instead of transferring data. We've all heard the new label "The World Wide Wait". Having the fastest modem on the block won't help you at all if there is a bottleneck somewhere beyond your local ISP, which of course is where we all want to go.
Well, just so you don't give up on the Internet entirely as a waste of time, let me tell you what has been going on behind the scenes to address this problem. After all, the same people who gave us the Internet (the military / university / academic community) are having to wait just like you and I. It is this collaboration of universities, government, and business which is again addressing the problem of providing high-speed Internet access to its members.
Before we get into the details let me put the problem into perspective. The Internet, as we know, began in the mid-sixties as a US military experiment. Universities and research centres were the major developers of the technology, and practically the only users. Universities are typically open to advanced networking technology, and are known as "early adopters". Transferring large chunks of data across the country with minimal waiting wasn't a problem, since the bandwidth far outweighed the usage.
The phenomenal growth of the Internet, spurred on by the World Wide Web and the introduction of commercial use, has reduced the bandwidth to a tiny trickle. Researchers can no longer utilize resources in the most efficient manner. The medical field, which has embraced the technology, uses the Internet for transferring X-ray files and so on; nuclear physicists need access to super-computers to perform complex data analysis and computations; both must wait while your e-mail goes through (there is no traffic prioritization on today's Internet).
First steps ...
Even early in the development of the World Wide Web it was possible to do multimedia conferencing without much trouble, but the Internet was getting noticeably slower. Along the way there have been various advances in networking technology to speed things up. One such advance was the introduction of a new version of the HTTP protocol, which is responsible for how web browsers interact with web servers. In version 1.0, every element of a web page required a separate connection to the server in order to make the download to your computer. That means that if you had a web page containing 20 images (be they buttons, balls, or large image files), your web browser had to make 21 separate connections: 1 for the text content of the page, and 1 for each image file (what a great way to inflate your "hit" count).
The latest version, 1.1, uses a "keep alive" feature, which lets a browser make one TCP/IP connection and transfer multiple files. The latest browsers partially support this protocol. This increases the speed of the connection considerably. However, as much as this development speeds things up, at the same time more and more users are coming online. A better answer was needed.
Internet2
In the US, the solution is Internet2, or I2, and is a research program of the Next Generation Internet (NGI). The project is handled by the University Corporation for Advanced Internet Development (UCAID). It's a consortium of about 120 US research universities, and will be a closed-loop Internet of just these universities. They will develop and test the new technologies, and then adopt the technology in the "old" Internet.
So far about 25 universities have been connected. The rates of speed we're talking are phenomenal: 2.5 billion bits per second! Currently the US backbone runs at 45 million bits per second (T3). The speeds involved in NGI are 100 to 1,000 times faster than today's Internet end-to-end. Throughout the backbone will be gigaPoPs (gigabit per second Points of Presence), which will act as hubs.
The Canadian Connection
The Internet backbone in Canada has been CA*net since 1990. It too has seen a shrinking of the pipe as more users come online demanding faster and more robust connections. In October 1996, board members of CA*net set in motion a plan to connect the universities, research centres, and private sector, with a next-generation Internet: CA*net II was born. Bell Advanced communications, a subsidiary of Bell Canada, assumed responsibility for the national backbone across Canada on April 1, 1997. In June 1997, CA*net II was launched to provide our universities and research communities the high-speed connections they need. It is a project of the Canadian Network for the Advancement of Research, Industry and Education (CANARIE) . CA*net no longer exists - it has been transformed into Bell's Internet Transit Service (ITS). A minor detail, since Bell has been operating the backbone since 1995.
Like its US counterpart, it is a collaborative effort of many parties: universities, major telecommunications carriers (Bell Canada, AT&T Canada, Teleglobe Canada, and the Stentor group). Equipment manufacturers Newbridge Networks and Nortel are also members. Again, the university and research community was chosen to develop the new network for the same reasons as the US initiative.
There are similarities between CA*net II and Internet 2, obviously, but one of the differences is significant: CA*net II intends to connect all of our universities rather than just some. In fact, we already have a working model called the National Test Network (NTN). It stretches 6,000 km from St. John's Newfoundland to Vancouver, British Columbia and is the world's largest high-speed test network. Along the way it connects 11 smaller test networks through 15 gigaPoPs. Each province has a Regional Advanced Network (RAN) through which a university can access the nearest gigaPoP, which will connect to the new backbone.
In addition, the NTN is connected to the high-speed European and Asian networks via Teleglobe's ATM (Asynchronous Transmission Mode) satellite. A connection to the US vBNS (very high-speed Broadband Network Service) is already in place. The vBNS is the technology supporting Internet 2.
You would be mistaken if you thought the new backbone will replace the old one. Each member is required to maintain its current connection to the Internet. The purpose of CA*net II is to handle traffic impossible to manage with the current network. This includes distance learning, telemedicine, and 3-dimensional virtual reality conferencing (tele-immersion). Several tests are currently being conducted using ATM technology. For example, the University of British Columbia is working with the forensic unit of the RCMP in Ottawa on a case involving scanning partial human remains. The images are transmitted over the network with diagnostic accuracy, and quite rapidly I might add. At the receiving end (the RCMP lab), the 2-dimensional images are used to create a 3-dimensional model as part of forensic reconstruction. Did I mention that the 3-dimensional model was being created as the 2-D images were being received!?
We end users will eventually see some of the benefit as well, or so the documentation reads. The idea is for the test beds to work out the bugs, then migrate the technology to the commercial sector, and finally end up at your keyboard. I can't wait. Imagine where the research will be by the time we get this batch! I've already seen references to CA*net III and the Optical Internet. The speeds involved here are in the range of 2.5 to 40 Gbps (Giga-bits per second) and better.
Want to find out more?
Here's a list of the resources I used to research this article.
Today: Thursday, 17-May-2012 21:50:41 EDT | Updated: Friday, 06-Aug-2010 14:43:19 EDT